Inter-domain electron transfer in cellobiose dehydrogenase: modulation by pH and divalent cations

The flavocytochrome cellobiose dehydrogenase (CDH) is secreted by wood-decomposing fungi, and is the only known extracellular enzyme with the characteristics of an electron transfer protein. Its proposed function is reduction of lytic polysaccharide mono-oxygenase for subsequent cellulose depolymerization. Electrons are transferred from FADH2 in the catalytic flavodehydrogenase domain of CDH to haem b in a mobile cytochrome domain, which acts as a mediator and transfers electrons towards the active site of lytic polysaccharide mono-oxygenase to activate oxygen. This vital role of the cytochrome domain is little understood, e.g. why do CDHs exhibit different pH optima and rates for inter-domain electron transfer (IET)? This study uses kinetic techniques and docking to assess the interaction of both domains and the resulting IET with regard to pH and ions. The results show that the reported elimination of IET at neutral or alkaline pH is caused by electrostatic repulsion, which prevents adoption of the closed conformation of CDH. Divalent alkali earth metal cations are shown to exert a bridging effect between the domains at concentrations of > 3 mm, thereby neutralizing electrostatic repulsion and increasing IET rates. The necessary high ion concentration, together with the docking results, show that this effect is not caused by specific cation binding sites, but by various clusters of Asp, Glu, Asn, Gln and the haem b propionate group at the domain interface. The results show that a closed conformation of both CDH domains is necessary for IET, but the closed conformation also increases the FAD reduction rate by an electron pulling effect.


Introduction
Cellobiose dehydrogenase (CDH; EC 1.1.99.18; CAZy database ID AA3-1) is detected in the secretome of wood-degrading, composting and plant pathogenic fungi. The results of recent studies support a role in the oxidative breakdown of crystalline cellulose or other polymeric, carbohydrate constituents of plant biomass in combination with lytic polysaccharide monooxygenase (CAZy database ID AA9) [1][2][3][4][5]. As a flavocytochrome, CDH exhibits a unique architecture for a secreted enzyme. It features an N-terminal electron-transferring cytochrome domain (CYT) that is bound to a carbohydrate-oxidizing dehydrogenase domain (DH). Electrons obtained from carbohydrate oxidation are transferred from the DH to the CYT by inter-domain electron transfer (IET). The proposed function of the CYT is to donate electrons to the copper centre of lytic polysaccharide mono-oxygenase to activate oxygen [2,6]. IET from FADH 2 to the b-type haem in the CYT is therefore an important step towards oxidative cellulose depolymerization.
Various pH optima for the IET have been reported in the literature. Usually, the activity with the in vitro electron acceptor cytochrome c (cyt c), which only interacts with the CYT [6][7][8], is used to approximately determine the IET. Direct observation of IET requires stopped-flow spectrophotometry, which has been performed for Phanerochaete chrysosporium CDH [9]. These studies show that CDHs from basidiomycetes (class I) have acidic pH optima, and IET ceases at approximately pH 6. The reported reason is electrostatic repulsion between the domains [10]. Ascomycetous CDHs (class II) are more diverse, and some of them exhibit less acidic or even alkaline IET pH optima [10,11]. This is remarkable because the reported isoelectric points for these CDHs do not differ from those of basidiomycetous CDHs with acidic pH optima. A possible explanation is that the DH and the CYT exhibit different patches of charged or noncharged amino acid residues.
The structure of full-length CDH is unknown, but structures for the isolated DH (PDB ID 1KDG) and CYT (PDB ID 1D7C) from P. chrysosporium are available [12,13]. Docking of the two domains has been reported, whereby a haem propionyl group was found to protrude into the substrate channel of the DH and come into close contact with the FAD [10]. The modelling is based on structural complementarity, and it is unclear whether this closed conformation is energetically favoured or whether the CYT is mobile and moves freely between a closed and an open conformation. The long linker (15 amino acids) between the domains suggests that the CYT has a high degree of freedom to move when not in contact with the DH. This appears to be necessary on order to transfer electrons to proteins such as cyt c or lytic polysaccharide mono-oxygenase. In previous stoppedflow measurements [9], no IET above pH 6 was observed for P. chrysosporium CDH, but the dehydrogenase domain was highly active at alkaline pH. One of two explanations may apply: (i) the domains separate because of electrostatic repulsion, and the increased distance between the domains prevents IET, or (ii) the closed conformation is stable and the domains are kept together, but IET is disrupted by structural and protonation changes along the electron transfer pathway. The possibility of reversal of the electron transfer by shifting the FAD redox potential to above the haem b potential has been excluded by Igarashi et al. [9].
A recent study revealed a way to investigate the pH dependency of the IET in more detail. It reported that calcium ions increased the enzymatic activity and catalytic current of three CDHs from Myriococcum thermophilum, Humicola insolens and Phanerochaete sordida [14]. Addition of a millimolar concentration of calcium ions to the buffer solution increased the catalytic current of CDH-modified spectroscopic graphite electrodes and also the turnover of cyt c. It was speculated that calcium ions modulate the domain interaction, and thereby enhance the IET between FAD and haem b cofactors. Here, we study the effect of the calcium concentration on twelve CDHs from various fungi to determine whether the effect is ubiquitous among CDH. We also investigate whether CDH is only IET-competent when both domains are in the closed conformation, and whether electrostatic repulsion above a certain pH prevents the closed state and therefore IET. M. thermophilum CDH was used to investigate the domain interaction and IET using spectroscopic, steady-state, pre-steady-state and calorimetric techniques. To elucidate the charges at the domain interfaces of three CDHs, comparative modelling and docking were performed.

Results and Discussion
Modulation of CDH activity by pH ions CDH from M. thermophilum (MtCDH), was previously reported to be responsive to calcium ions [14]. A millimolar concentration of calcium ions increased the activity of the enzyme fivefold. Using a screening approach based on artificial electron acceptors for MtCDH, we also observed such enhancement when using other divalent earth alkaline metals such as Mg 2+ , Sr 2+ , Ba 2+ and Cd 2+ (Fig. 1), but not when using monovalent cations or anionic species (Table  S1). The effect is stronger at alkaline pH, and is only observed for CDH activity towards cyt c, and to a much lesser extent for reduction of the two-electron acceptor 2,6-dichloroindophenol (DCIP). Reduction of the two-electron acceptor 1,4-benzoquinone (BQ) was not affected at all. Cyt c is solely reduced by the haem b of CDH and requires IET. The two-electron acceptors DCIP and BQ are reduced directly by FADH 2 and do not require IET and the cytochrome domain. The increase in electron acceptor turnover in the presence of divalent cations depends on their concentration. Below a 3 mM concentration, only a small increase in cyt c turnover by CDH was detected. The maximal effect was found at concentrations between 30 and 60 mM, while higher concentrations also decreased the turnover of cyt c (Fig. 1A). A smaller increase was found for DCIP turnover, where the optimum cation concentration was also 30-60 mM, but no decrease at higher concentrations was observed (Fig. 1B). Apparent catalytic constants of MtCDH for cyt c in the presence of 30 mM of divalent cations were measured at pH 5.5 and 7.5, and compared to a 30 mM concentration of Na + ( Table 1). The increase in k cat caused by divalent cations was 3.5-4-fold at pH 5.5, but 160-200-fold at pH 7.5. The huge enhancement at pH 7.5 is only partly attributable to an increase in the absolute turnover of cyt c by CDH (1.2-1.4-fold higher compared to pH 5.5), and is mostly due to a reduced of cyt c turnover in the absence of Ca 2+ . At pH 7.5 and in the absence of divalent cations (Na + was used as substitute), cyt c turnover is negligible. In the presence of divalent cations, the catalytic efficiency for cyt c is very similar at pH 5.5 (mean of~2.8 9 10 5 M À1 Ás À1 ) and pH 7.5 (mean of~3.2 9 10 5 M À1 Ás À1 ). This indicates that divalent cations strongly counteract the shutdown of the IET at neutral/alkaline pH, and do not influence electron transfer from the CYT to cyt c. Apparently, neither the atomic radius (Ba 2+ > Sr 2+ > Ca 2+ > Mg 2+ ) nor the electronegativity (Mg 2+ > Ca 2+ > Sr 2+ > Ba 2+ ) of the earth alkali metal ions significantly modulate the observed enhancing effect. A lower enhancing effect was found for Cd 2+ , which increases cyt c turnover at pH 7.5 to 0.15 s À1 , compared with a turnover of 1.3 s À1 observed in the presence of Ca 2+ (Fig. 1A).

Modulation of activity in CDH from other sources
Twelve CDHs from 11 fungi were subjected to screening using the three electron acceptors cyt c, DCIP and BQ at three pH values in the presence or absence of     (Tables 2 and 3). None of the investigated enzymes showed activation of BQ turnover in the presence of Ca 2+ . DCIP turnover was slightly increased for some class II CDHs at pH 7.5. However, the turnover of cyt c was increased for both CDH classes, typically 1.4-fold for class I and between 1.3and 4.2-fold for class II. At higher pH, the observed increase was most often higher,~2-4-fold for class I CDHs and 3-5-fold for class II CDHs. In contrast to ascomycetous class II CDHs, which show higher pH optima, the measured pH for class I CDHs was pH 6.5, and no activity for cyt c was detected at pH 7.5. Interestingly, there are exceptions from the observed increase in activity: The class I CDH from Sclerotium rolfsii showed a decreased cyt c turnover at pH 6.5, and the class II Neurospora crassa CDH IIA showed a decreased cyt c turnover at pH 5.5 and 7.5 in the presence of Ca 2+ . Whether these enzymes are inert to the presence of divalent cations, or the pH chosen was not suitable, cannot currently be determined. The 125-fold activation of MtCDH activity at pH 7.5 in the presence of Ca 2+ is very high compared to the other investigated CDHs. Despite these excep-tions, the enhancement effect of divalent metal cations was observed for all tested CDH enzymes.

Modulation of electron acceptor pH profiles by Ca 2+
The observed pH dependency of the electron acceptor turnover in the presence of divalent cations was studied in detail using MtCDH and Ca 2+ partes pro toto (Fig. 2). The turnover rate of the positively charged one-electron acceptor cyt c by the CYT depends on the IET, and is optimum at pH 5. Above this pH, the turnover of cyt c decreases rapidly, and disappears at pH 6.5. As the charge of cyt c (pI 10.0-10.4) is not altered at this pH, the change in turnover is probably attributable to reduction of the IET between the DH and the CYT. At pH 5, the side-chain carboxy functions of surface-exposed aspartic acid (pK a = 3.86) and glutamic acid (pK a = 4.07) are~90% deprotonated, creating a strong electrostatic repulsion between both domains. In the presence of Ca 2+ , this reduction of turnover above pH 5 is not observed, indicating that the electrostatic repulsion is neutralized. The increasing turnover of cyt c  at increasing pH shows that the reductive cycle at the FAD has an alkaline pH optimum. The positively charged electron acceptor ferrocenium showed similar behaviour in the absence or presence of Ca 2+ Increasing turnover at alkaline pH indicates that this one-electron acceptor does not depend on IET, but is reduced by the FADH 2 directly. The activity of CDH towards the two-electron acceptors DCIP (with a strong negative partial charge) and BQ (with weak negative partial charge) was slightly increased in the presence of Ca 2+ at pH values above 6. Interestingly, this activation was not observed for the single DH, which indicates that not only do the negative charges around the substrate channel reduce the accessibility for electron acceptors, but that the proximity of the haem b also affects electron acceptor-mediated oxidation of FAD.
The most interesting behaviour was observed for the ferricyanide anion. This one-electron acceptor may be directly reduced by FADH 2 in the DH, but only at pH values below 5.0. Such behaviour was also observed for glucose oxidase [15]. For CDH, ferricyanide reduction may be observed up to pH 6.0, demonstrating that the proximity of the CYT to the DH influences ferricyanide reduction more strongly than it influences DCIP or BQ reduction. The pH profiles of the DH in the presence and absence of Ca 2+ reveal that the charge-induced reduced accessibility of the ferricyanide anion to the active site is a minor factor. The increasing ferricyanide turnover by CDH at alkaline pH in the presence of Ca 2+ may be explained by two models. In model 1, the close proximity of the haem b to the FAD increases ferricyanide turnover at FAD. In model 2, the reduction of ferricyanide occurs at the CYT above pH 5. The two pH optima observed in the pH profile promote the second mechanism. The breakdown of ferricyanide and cyt c reduction above pH 6.0-6.5 provides a measure for shutdown of the IET. If the pH increases above pH 6.5, the CYT of MtCDH no longer interacts with the DH, because electrostatic repulsion of the negatively charged domains prevents adoption of a closed-state conformation ( Table 2, pI values). In the presence of Ca 2+ , the ferricyanide turnover of CDH shows a monotonic increase from pH 5-8. The pH profiles of ferricyanide and cyt c therefore give a clear indication that divalent cations support adoption of a closed conformation by the DH and the CYT when the domain interface becomes negatively charged at neutral/alkaline pH. This effect is not caused by the presence of distinct binding sites for divalent cations, as these were not found in our docking models of the DH and the CYT, and the 30-60 mM concentration of dications required to achieve the optimal effect is magnitudes higher than the reported affinities for Ca 2+ specific EF-hand binding sites, for example (2.1 lM [16] or 0.49 lM [17]). The high concentration required suggests unspecific shielding of close opposing charges at both sides of the domain interface. Similar behaviour has been reported for surfactants, DNA and hyaluronan in the presence of Ca 2+ , and was termed the 'divalent cation bridging effect' [18][19][20][21][22].

Spectral features of MtCDH
Electron absorption spectra of oxidized MtCDH in the presence of NaCl or CaCl 2 were recorded at pH 7.5, and showed the typical Soret-band maximum at 421 nm for haem b (Fig. 3A). Upon reduction, the absorbance of the haem a and b peaks at 533 and 563 nm increased, the Soret band shifted from 421 to 429 nm, and the FAD absorbance at 450 nm decreased. Addition of calcium chloride to a final concentration of 30 mM had no effect on the spectra of the oxidized and reduced enzyme. Measurements at pH 5.5 showed the same result. No indications of Ca 2+ -induced changes in the coordination environment of the haem b cofactor were found.
Inter-domain electron transfer of MtCDH in the presence of CaCl 2 Based on the spectral characteristics of CDH, stoppedflow spectroscopy was used to monitor reduction of the FAD and haem b cofactors of CDH. Rapid mixing of MtCDH (5 lM final concentration) with an excess of cellobiose (3 mM) results in fast FAD reduction, followed by IET and one-electron reduction of the haem b. For comparison, the dehydrogenase domain from M. thermophilum (MtDH) was also analysed. Figure 3B shows apparent pseudo first-order rate constants (k obs ) for both reduction steps between pH 4.5 and 8.5. The FAD reduction changes with pH, and shows a maximum of 38.5 s À1 at pH 8.5 in the presence of 30 mM NaCl. In the presence of 30 mM CaCl 2 , k obs is only slightly higher and reaches a maximum of 41.6 s À1 at pH 7.5. These differences are negligible compared to the changes Ca 2+ exerts on full-length CDH. The FAD reduction rate in the absence of the dication is almost identical to that in the DH, but in the presence of Ca 2+ , the k obs increases 1.3-1.7-fold. The close proximity of the haem b appears to exert a pull effect on the FAD, resulting in a faster FAD reduction rate. From docking experiments, a mean edge-to-edge distance of 8 AE 2 A between FAD and the haem b proprionate chain was calculated, which is close enough for fast electron transfer. Modulation of the FAD redox potential by electronic coupling through the bridging Trp295 residue may explain the increased steady-state turnover of DCIP and BQ in the presence of Ca 2+ (Fig. 2E,G and Table 2). Both pre-steady-state and steady-state experiments indicate an alkaline pH optimum of approximately pH 8 for the reductive half-reaction at the FAD, which differs greatly from the acidic pH optimum (4.5-5.0) observed for P. chrysosporium CDH [9].
The pH-dependent interaction of both CDH domains shows maximal reduction of haem b and therefore IET at pH 4.5 (Fig. 3B). The pH optimum of PcCDH is slightly lower (pH 3.5 [9]). In agreement with the steady-state measurements using cyt c, an effect of calcium ions was also observed here. Apparent haem reduction rates at pH 4.5 increased from 0.25 to 1.37 s À1 upon addition of CaCl 2 . At pH 6.5 and above, the haem reduction became extremely slow. At pH 7.5, only~5% of haem b cofactors was reduced within the observed time span of 30 s, but haem reduction was recovered by addition of Ca 2+ . In the presence of Ca 2+ , k obs decreases monotonically from pH 4.5 to 7.5, but the rate is always higher compared to the rates measured in the absence of Ca 2+ . This decrease indicates that, with increasing pH, the cation-bridging effect is overcome by electrostatic repulsion. A comparison with the steady-state cyt c pH profile shows that the positive charges of this proteinogenic electron acceptor strongly influence the IET. Turnover maxima of MtCDH for cyt c are found at pH 5 in the absence of Ca 2+ and at pH 8 in the presence of Ca 2+ , whereas the IET measured by stopped-flow in the absence of cyt c continuously decreases from its maximum at pH 4.5, in either the presence or absence of Ca 2+ . These results indicate the limits of the use of cyt c to measure IET.

Effect of calcium chloride on the thermostability of MtCDH
Transition mid-point temperatures (T M ) of MtCDH in the presence of 30 mM NaCl or 30 mM CaCl 2 were recorded at pH 5.5 and pH 7.5. At pH 5.5, overlapping peaks were observed, with a maximum (T M ) at 68.5°C (Fig. 3C and Table 4). At pH 7.5, the presence of CaCl 2 shifted the thermostability only slightly by 0.7°C (from 70.7 to 71.4°C). The observation of a slightly higher T M at pH 7.5 shows that the closed conformation at pH 5.5 does not lead to higher thermal stability of the enzyme, which suggests that no strong interactions are involved in forming the closed structure. It also confirms that stabilization of the closed conformation of CDH in the presence of Ca 2+ is achieved by weak forces rather than strong binding. To further test whether divalent ions exert a permanent interaction with MtCDH, the enzyme was incubated in 30 mM CaCl 2 at 22°C for 2 h. Dilution in a buffer containing 30 mM NaCl showed that the activity- enhancing effect disappeared and the activity became similar to that of the negative control incubated in buffer containing 30 mM NaCl (Fig. 4A). Incubation of the enzyme in 30 mM NaCl and subsequent measurement of cyt c activity in the presence of 30 mM CaCl 2 , resulted in the same enhancement of activity. Based on the results of these experiments, it may be concluded that the nature of the interaction is transient and does not increase the stability of CDH. To exclude potential interferences from residual divalent metal ions in the CDH preparations, CDHs from Phanerochaete sordida, Corynascus thermophilus and Myriococcum thermophilum (0.2 mgÁmL À1 ) were incubated with 5 mM EDTA for 1 h at 22°C. After incubation, enzymes were extensively diafiltered using EDTA-free buffer. The same procedure was applied to cyt c. The cyt c activities of the EDTA-treated and untreated enzymes were measured, and no difference was observed (Fig. 4B), indicating that no interfering ions were present.

Structure analysis and docking of CDH domains
The characterized MtCDH, together with CDH from P. chrysosporium (PcCDH) and C. thermophilus PcCDH and MtCDH show the highest number of negatively charged amino acid residues in the interfacial area (five for PcCYT, ten for PcDH, seven for MtCYT and eight for MtDH), whereas CtCDH has fewer charges (five for CtCYT and seven for CtDH). However, the number of charged amino acids alone does little to explain the domain behaviour and IET of the selected CDHs. Therefore, titration plots of the surface-exposed amino acids in the interfacial area were created (Fig. 5). The calculated isoelectric points are similar for PcCDH and MtCDH (~4.0 for the CYT,~4.5 for the DH), consistent with experimental data that show the same upper pH limit of IET for both enzymes (6). The titration plots also correctly predict a higher pH for electrostatic repulsion of the domains for CtCDH (pI of~5.0 for the CYT and 6.2 for the DH), which does indeed show a higher pH for its IET optimum. However, the two-dimensional view does not give information on the divalent cation-bridging effect. Therefore, the interface of the docking models was investigated. The obtained docking positions do not differ by much, but all models were evaluated, and the best was selected on the basis of (i) a close haem b to FAD distance, and (ii) a suitable length and orientation of the linker peptide. Interestingly, the distance between the haem b propionate A B  group reaching into the substrate channel and the FAD isoalloxazine ring (edge-to-edge distance) was shortest for CtCDH (5.6 AE 0.6 A) compared to MtCDH and PcCDH (8 AE 2 and 8.6 AE 2 A, respectively). A closer distance between the cofactors in the closed confirmation may additionally support the occurrence of IET at an increased pH under which the domains begin to separate. The mean calculated interfacial area between the DH and the CYT for the three CDHs differs by a maximum of 16% (MtCDH 2270 AE 280 A 2 , CtCDH 2090 AE 70 A 2 , PcCDH 1910 AE 250 A 2 ). Docked structures with highlighted cofactors and distances below 13 A between opposing Asp, Glu, Asn and Gln residues are shown in Fig. 5. Such clusters, which inadequately mimic the structurally well-defined Ca-binding sites in calmodulin, for example, allow weak electrostatic interactions between Ca 2+ and negatively charged amino acid side chains. It is obvious that, in MtCDH, these clusters contain many residues, but in PcCDH and CtCDH, they contain only a few. Four clusters were found in MtCDH and PcCDH, and three in CtCDH (Table S2). For all enzymes, at least one cluster involves one or both haem propionate groups. The much higher number of potential dication ligands in MtCDH provides an explanation for the high increase of IET at alkaline pH. The presence of more sites for Ca 2+ binding more efficiently counteracts electrostatic repulsion of the domains. Interestingly, for a number of proteins crystallized in the presence of cadmium ions, it has been observed that Cd 2+ is positioned at the interface between two neighboring protein molecules and coordinated by two carboxyl groups of glutamic or aspartic acid side chains each belonging to one of the two molecules [23][24][25]. This coordination may be supplemented by interactions with carbonyl groups of the polypeptide backbone. Such complexation was also observed in the structure of P. chrysosporium CYT (PDB ID 1D7C [13]). Either one or both of the Cd 2 + -complexing sites known for dimers of P. chrysosporium CYT (Fig. 5D) were found in the predictions for PcCDH, MtCDH and CtCDH.

Conclusions
Steady-state and pre-steady-state measurements in combination with homology models show that the pH dependence of IET and the reduction of cyt c may be explained by the presence of negatively charged amino acids at the interface of the DH and CYT, preventing formation of a closed conformation of the DH and the CYT. Calculated titration plots of these interfaces demonstrate that the CYT has a lower pI than the DH; the same trend was also observed for the isoelectric points of the domains [10]. Below, or a little above these pIs, the domains do not repulse each other, and thus allow electron transfer between the domains.
In the present work, we also unequivocally demonstrate that dications enhance the IET reaction in cellobiose dehydrogenase, as first suggested by Schulz et al. [14]. Catalytic constants for cyt c at pH 7.5 measured in the presence of calcium ions indicate that the presence of divalent cations supports a closed conformation of both domains at neutral and alkaline pH. The reduction state of the cofactors during catalysis, monitored by stopped-flow spectroscopy, reveals that Ca 2+ affects the IET most strongly. Less prominent, but clearly observable, is the coupling of both cofactors in close proximity of the domains, which increases the reduction rate of FAD. This effect occurs either at low pH or in the presence of Ca 2+ . Currently, we have no supported explanation for this phenomenon, which was observed in steady-state and pre-steady-state kinetic measurements, but the observed pull effect on the FAD may originate from a modified redox potential of the FAD via an intermediary Trp295 side chain.
It has been reported [18][19][20][21][22] that divalent cations potentially form a bridge between oppositely charged carboxylate groups on polyelectrolytes, DNA or hyaluronic acid. Based on our data, this effect also influences domain interaction in CDH, and presumably also applies to other proteins, given that a number of negatively charged amino acid residues are present in close proximity at domain interfaces. In addition to providing an explanation for the observed pH-dependent IET of CDH and the Ca 2+ ion effect, these findings may also have practical relevance, e.g. for increasing the current output of CDH-based biosensors or biofuel cells. Also, in accordance with the results of a previous study [26], a CDH-based bioelectrochemical switch may be accomplished at neutral/ alkaline pH using divalent cations to switch the IET and hence direct electron transfer on and off. The high millimolar dication concentrations required for such an effect are rarely encountered in nature, but may provide ways to stabilize or regulate enzymes in biocatalytic and biomedical applications.

Steady-state kinetics
CDH activities were assayed spectrophotometrically based on the cellobiose-dependent reduction of the two-electron acceptors DCIP (0.3 mM, e 520 = 6.9 mM À1 Ácm À1 ) and BQ (1 mM, e 290 = 2.24 mM À1 Ácm À1 ) and the one-electron acceptors cyt c (0.02 mM, e 550 = 19.6 mM À1 Ácm À1 ), potassium ferricyanide (1 mM, e 420 = 0.98 mM À1 Ácm À1 ) and ferrocenium hexafluorophosphate (0.1 mM, e 300 = 4.3 mM À1 Ácm À1 ). All assays contained 3 mM cellobiose, and were performed in 50 mM sodium formiate buffer (pH 3-4), 50 mM sodium acetate buffer (pH 4-6) or 50 mM MOPS buffer (pH 6-8.5). Reactions were started by mixing 20 lL of enzyme solution (0.05-0.2 UÁmL À1 ) with 980 lL of a pre-warmed reaction mix containing the electron acceptors and cellobiose. Reduction of electron acceptors was followed for 180 s at 30°C in a Lambda 35 spectrophotometer (Perkin Elmer, Waltham, MA, USA) with a thermo-controlled eight-cell changer. Enzyme activity is defined as the amount of enzyme that reduces 1 lmol of the respective electron acceptor per minute under the specified conditions. Catalytic constants were calculated from initial rates by using non-linear least-squares regression to fit the observed data to the Michaelis-Menten equation using SIGMA PLOT 11 (Systat Software, San Jose, CA, USA). The influence of various anions and cations on CDH activity was measured in 96-well plates in an EnSpire multimode plate reader (Perkin Elmer). For measurements below 340 nm, UV-transparent 96-well-plates or UV-transparent cuvettes were used. Protein concentrations were determined by the dye-binding method [32] using a pre-fabricated assay (Bio-Rad, Hercules, CA, USA) with BSA as the calibration standard. Concentrations of MtCDH used for stopped-flow spectroscopy were determined based on the enzyme's absorption coefficient at 421 nm (e 420 = 99.4 mM À1 Ácm À1 ).

Pre-steady-state kinetics
Fast kinetic studies were performed on a SX-18MV spectrophotometer (Applied Photophysics, Leatherhead, UK). The cellobiose-dependent reduction of CDH was monitored in single mixing mode using a SX/PDA photodiode array detector or a SX/PMT photomultiplier tube. The reduction of FAD was followed at 449 nm and the reduction of haem b was followed at 563 nm. Concentrations of CDH and cellobiose after mixing were 5 lM and 3 mM, respectively. All measurements were performed at 30°C with at least five repeats for each investigated condition. Observed rate constants (k obs ) were calculated by fitting the absorbance changes to a double exponential curve using PRO-DATA software (Applied Photophysics).

Differential scanning calorimetry
The transition mid-point temperature (T M ) of MtCDH in the presence of CaCl 2 and NaCl was determined using a MicroCal VP-DSC calorimeter equipped with an autosampler (MicroCal, Northampton, MA, USA). MtCDH was adjusted to a concentration of 1 mgÁmL À1 based on its molar absorption at 421 nm. Thermograms were obtained at pH 4.5 (100 mM sodium acetate buffer) and pH 7.5 (100 mM MOPS buffer) between 30°C and 100°C at a scan speed of 1°C min À1 . Heat-inactivated enzymes were re-scanned as a control, and their values were subtracted from the experimental thermograms. Data were evaluated using ORIGIN 7.5 software (Origin Lab Corporation, Northampton, MA, USA).

Comparative modelling and docking
Structure-guided homology models of the individual DH and CYT of MtCDH (GenBank accession number ABS45567.2) and CtCDH (GenBank accession number ADT70772.1) were generated using the SWISS-MODEL server (http://swissmodel.expasy.org/) [33] with the individually crystallized DH of P. chrysosporium CDH (PDB ID 1KDG [12]) and the CYT of P. chrysosporium CDH (PDB ID 1D7C [13]) as templates. The HADDOCK webserver (http://haddock.science.uu.nl/) was used for protein-protein docking of the individual CDH domains [34]. To roughly pre-define the docking position, residues located within 4 A of the haem b CYT (Pro72, Tyr98, M100, Gln173, Gln174, His175 and Met179) and residues in proximity to the substrate channel in the DH (Met80, Ala81 and Val91) were selected as initial docking sites. In total, 12 models from three clusters per enzyme were evaluated. Models with a haem b to FAD distance of more than 10 A were not considered for further analysis. Structures were visualized using the PyMOL MOLECULAR GRAPHICS SYSTEM, version 1.4 (Schr€ odinger, New York, NY, USA).

Supporting information
Additional supporting information may be found in the online version of this article at the publisher's web site: Table S1. Influence of anions and cations (30 mM) on MtCDH activity measured using cyt c, DCIP and 1,4benzoquinone. Table S2. Distances between Asp, Glu, Gln and Asn residues in the docking models.